Lung adenocarcinoma (LUAD) is the most histologically diverse lung cancer subtype and frequently metastasizes to the central nervous system (CNS). Our goal is to identify the molecular determinants of LUAD lineage and histological heterogeneity and their biological consequences for metastasis. We previously identified a novel lineage selective transcriptional program that constrains LUAD cell invasion, and showed that this pathway is suppressed in high grade LUADs. Here, we provide evidence that invasive LUAD cells can activate a neuroendocrine-like gene signature, which is further induced once tumor cells disseminate into the brain. This phenomenon is linked to pervasive chromatin alterations in KRAS mutant cells that are competent for metastasis. Moreover, activation of neuroendocrine-like genes in human tumors correlates with distinct morphological subgroups of LUAD at risk for CNS relapse. We hypothesize that: 1) LUAD lineage reprogramming is driven by chromatin modifying proteins, 2) histological heterogeneity is a measure of the adaptive capacity of high grade LUADs, and 3) the activation of the neuroendocrine lineage in particular correlates with their predisposition for CNS relapse. Many neuroendocrine genes encode for secreted proteins with dual glandular and neuroactive functions. Hence, we also propose that 4) a subset of neuroendocrine-like genes enables disseminated LUAD cells to establish a neurogenic niche that is required for brain metastatic outgrowth. Our mechanistic predictions will be studied by integrating epigenomics, functional genomics, and experimental biology. These methods are applied to existing models of LUAD, as well as new models, which are engineered in mice or derived from biopsies of patients consented through our unique Yale Lung Cancer Patient Biopsy protocol. In Aim 1, we performed a functional genomic screen to nominate the histone methyltransferase ASH1L as a novel epigenetic driver of lineage plasticity and metastatic competence. We will ascertain the stage specific requirement for ASH1L during LUAD progression and metastasis, and identify the mechanism by which it regulates the transcriptome of metastatic cells. In Aim 2, we identified the dual neuroendocrine/neuroactive factor FGF9 as being epigenetically activated in LUAD cells and required for brain metastasis. We will genetically test the prediction that tumor cell derived FGF9 stimulates oligodendrocyte progenitor cells to support metastatic outgrowth via paracrine Sonic Hedgehog signaling. Finally, under both Aims 1 and 2, we will utilize human biospecimens to study the relationship between epigenomic alterations, neuroendocrine marker expression, histological heterogeneity, and clinical outcome in LUAD. Thoracic malignancies account for most cancer-related deaths. Our complementary Aims provide a cogent mechanistic framework to understand the biological link between pulmonary specification, neurogenic functions, and CNS relapse. Finally, our proposal will generate significant insight as to how prospective epigenetic therapies can be harnessed for adjuvant therapy and/or the treatment of late stage brain metastasis.